Every brief here started as somebody complaining in public. Shipbrief keeps the quotes, counts how often the complaint comes back, and writes up what you would actually have to build.
39 briefs published, each one linked back to the posts it came from.
5 briefs
In Jan 2026 curl, tldraw, and Ghostty all instituted hard bans on AI-generated contributions because LLM-spam PRs and fake CVE reports are now a primary maintainer-burnout vector. Daniel Stenberg (curl): 'Fake and otherwise made up security problems effectively prevent us from doing real project work and waste time and resources.' GitHub provides zero native filtering.
Supabase users repeatedly hit a wall where local supabase/migrations stops matching what's actually applied in the remote Postgres. The tool then refuses to push, refuses to pull cleanly, and offers no first-party 'baseline + reconcile' path. Users end up doing manual SQL surgery on prod or abandoning the migration system entirely. From farmhutsoftwareteam: 'We cannot sync existing production databases when local migration files don't match the 141 migrations already applied remotely.'
Teams struggle to centralize requests scattered across GitHub issues, Discord messages, and Slack threads into one task system, causing duplicated effort and missed handoffs. One builder extended their tool with GitHub/Discord sync precisely because transferring requests to a single place was a recurring pain.
Builders report that connecting AI agents to real SaaS apps is the painful part — not calling one API, but auth, action naming, app objects, and glue code between tools. Agents also stall at login screens and at systems with no API at all, like government portals and banking sites.
Indie hackers collecting product feedback on social media have no seamless way to turn tweets into actionable development tasks; one developer explicitly wished for a tool to convert tweets into GitHub issues. Feedback ends up scattered across DMs, replies, and screenshots, and feature requests get lost.
A scraper reads public posts on X, Reddit and Hacker News every day — the ones where somebody is describing a problem rather than launching something.
Claude pulls the specific complaint out of each post. The same complaint, from different people in different places, collapses into one node in a graph that keeps growing.
Once a cluster has enough corroboration it becomes a brief: the problem, who has it, who already tried to solve it, and what you would have to build. Nothing is published off a single post.